一种改进的粒子滤波算法在SINS初始对准中的应用

摘要

在实际工程环境中，针对捷联惯导系统（SINS）大失准角初始对准中噪声统计特性未知的问题，设计了一种基于H滤波算法的鲁棒无迹粒子滤波算法（RUPF）。通过将无迹卡尔曼滤波算法（UKF）和鲁棒环节引入到粒子滤波(PF)的重要性密度函数中，得到了 RUPF 算法，提高了算法的鲁棒性。通过半物理实验，将 RUPF 算法与无迹粒子滤波算法（UPF）在 SINS 静基座大失准角对准中的性能进行了比较，在不同实验条件下，航向失准角精度至少提高了40%，对准精度优于0.05°，对准时间减少了约50 s。实验结果表明，RUPF 算法可以以较高的精度和较快的速度完成大失准角初始对准，且对准精度和对准速度均优于UPF算法。%In real engineering environments, the noise statistical characteristics are unknown in the initial alignment of SINS with large misalignment angle. To solve this problem, a RUPF algorithm is designed based onHfiltering algorithm. By combining UKF algorithm and robust link into importance density function in PF, the RUPF algorithm is obtained to improve the robustness of this algorithm. By means of emi-physical experiment, the filter performance of RUPF and UPF in SINS initial alignment on a static base is compared with that of large misalignment angles under various experimental conditions, which show that the accuracy of heading misalignment is increased by at least 40%, the alignment accuracy is better than 0.05°, and the alignment time is reduced about 50s. These results show that the RUPF can realize the initial alignment of SINS with large misalignment angles, whose alignment accuracy and alignment speed are higher than those of UPF.